The evolution of information technologies has currently reached the point allowing us to state the fact of formation of a new type of society, extending further on from informational to digital. The pivotal role in its formation belongs to virtual communications based on technical and software tools created by man. Novel channels of virtual communication greatly surpass the natural forms of communication in terms of the amount of transmitted information, which significantly affect the processes of socialization and orientation of a person under present conditions. Among important phenomena embodying the features of informational, virtual processes characteristic for the 21st century, are Internet memes, representing information units in the modern world. Internet memes in a virtual space accumulate a significant potential for influencing public opinion. In the paper, we overview the procedure for generating a query for a Google search engine in the context of designing an intellectualized monitoring system for Internet memes. The procedure of a search query generation is implemented based on the analysis of the two information sources: the first one containing data on relevant sociopolitical agenda and the second one - on the public perception and interpretation of this information. The proposed procedure of query generation has certain advantages in that it can easily be translated into a relevant software code and takes into account the dynamics of social and political processes in Russia.
Modern-day communications have largely moved to a virtual environment. The latter still awaits its proper examination because such communications and their content greatly affect undergoing social processes.
Internet memes belong to a form of such communications. The overall quantity of
Internet memes, which can be defined as semi-structured datasets, keeps rising steadily
[
What we deal herewith is an open and constantly replenishing system, not only in
terms of its content, but also in terms of genre diversity (slogans and Photoshop-memes,
viral videos and parodies, melodies and gifs, comic strips and demotivators), and
semiotic characteristics (verbal (text), auditory, visual and mixed messages) [
Internet memes in the virtual place possess a big potential to influence public
opinion. Hence the importance of producing algorithms (to be then implemented in software
products) designed to monitor the processes that determine their origin, structure, and
distribution in an automated mode. Among other things, this is because processing the
data stream of Internet memes, that is, the big data analytics can be hardly conducted
by people on their own. Too much information is needed to get processed and
comprehended [
Possessing a viral nature of distribution, Internet memes not only “infect”
communicants, but also change their thoughts, attitudes, and behavior [
In the paper, we tackle the procedure for generating a query, taken as a component of the intellectualized system for monitoring Internet memes.
Supported by the Russian Foundation for Basic Research (№ 20-011-31460), the project "Political meme in virtual communications: software product design for monitoring its distribution and influence (analysis of the Russian-speaking Internet segment)" has also the aim, among other things, to make a primary database of political Internet memes.
The paper’s goal is to describe the procedure of formulating a search query necessary for creating the database for political Internet meme images.
When formulating the search query aimed at building up the primary database named “The database of political Internet meme images” (IDB), it was principal for us to design such a mechanism that could be easily translated into computer software (SW). Our task was to find a constantly and automatically updated source of information reflecting relevant social and political processes in Russia. 2
When formulating the search query, we used the two information sources complying with the noted requirements. The first source of information included the headlines for the news of sociopolitical nature over a certain period taken from the newspaper Kommersant (Businessman). In the fully functional version, when formulating queries, feeds from news aggregators will be analyzed. A set of news headlines collected over some time forms the hypertext.
The second source contained the data from the weekly opinion poll “FOMnibus”
(Public Opinion Foundation). In particular, it was analyzed an array of data extracted
from respondents' answers to an open question: "What events of the past week, which
were reported in mass media, seemed most interesting to you or attracted your
attention?" [
The query itself consists of two parts. The first one is to reflect and define the subject
of a judgment (query predicate). Within this work’s framework, the predicate will be
defined as a constant term, namely, meme. The second part is variable and reflects the
studied subject in the structure of virtual communication. To receive search results
closest to the content of the query, the query wording is placed in quotation marks (“meme
…”) [
To fulfill the study, the Google search engine will be used, as, according to many objective indicators, it addresses the most complete database of images, while the algorithm of query making appears to be easy to get formalized. 3
To test the proposed approach, let us set the period as follows: from March 01 to March 31, 2020.
Based on the news feed archive from the Kommersant newspaper [
Fig. 1. The semantic network built for the hypertext relevant to the period from March 01 to
March 31, 2020.
As a next step, let us go up to the analysis of the constructed graphs, the aggregate of which defines the semantic network of the hypertext for the given period. To describe the principle of word selection when formulating a query, let us consider two graphs, whose choice is arbitrary. Figures 2 and 3 show the examples of graphs built for the semantic network of the hypertext. Let us denote them according to the names of the corresponding graph vertices: “Coronavirus” (graph in Figure 2) and “Regime” (graph in Figure 3). Fig. 2. Graph «Coronavirus».
The query is composed of words associated with the vertex (word) having the highest degree in the graph. To do so, we build a ranked sequence of degrees for the graph’s vertices (Table 1, Table 2). The query includes only the words (vertices) adjacent to the vertex having the greatest degree.
Based on the data, shown in Table 1, we can formulate the following queries: “Coronavirus new case infection Russia”; “Coronavirus new case infection die”; “Coronavirus new case infection fight”.
Based on the data, shown in Table 2, we can formulate the following queries: “Regime introduce emergency self-isolation”; “Regime emergency introduce self-isolation”; “Regime introduce emergency Moscow”; “Regime emergency introduce Moscow”.
Table 3 shows the query wordings and image search frequency results when the Google search engine was used. Found images can be downloaded by using the Download All Images tool. Next, the collected images were examined to eliminate duplicates with the help of the Find.Same.Images.OK application. This application seeks duplicate images by comparing them at the pixel level. This allows for efficient comparative analysis and filtering of the same images, even after they have been edited (rotated, mirrored, cropped, etc.). As a result, based on 1119 images, we accumulated 525 unique sets of queries.
For the whole four weeks in March 2020, the first place in the ranking of the most important events, according to the opinion poll "FOMnibus", was occupied by the Coronavirus epidemic, Coronavirus pandemic. Based on all the respondents’ answers, specific semantic units were selected, then formulated in the form of search queries. Table 4 shows a set of queries and frequencies of their use when searched in the Google search engine. The primary collection of these queries consisted of 2,699 images. On fulfilling the procedure of comparison and elimination of duplicates, 1401 images remained.
Query coronavirus monitored virus from China coronavirus spreading Europe Coronavirus tired of artificially made epidemic of Coronavirus Coronavirus, pandemic disease terrible, Coronavirus Coronavirus, phobia disease the threat of Coronavirus infection Coronavirus, widespread in Russia Coronavirus: soon quarantines everywhere virus – panic closure of borders school quarantines psychosis pumped up due to Coronavirus Coronavirus, artificially created panic Total
Judging by the data in Table 5, extending the structure of search queries (based on the Source 1, we generated 7 queries and based on the Source 2 – 16 queries) resulted, on the one hand, with a quite obvious consequence – an increase in the number of search results (images), and, on the other hand, an increase in the number of the unique search results. To sum up, the structure of queries makes an impact on the statistics of search results. 4
Having aggregated the arrays of images collected through the use of the queries formulated with the help of two sources, we obtained 1926 images. The procedure of their comparison reduced their number down to 1791 items. As a result, we created an IDB ready for further analysis and defined the mechanism of formulating queries.
This database requires to be first filtered according to certain objective criteria, which must be specified and formally described. This is necessary to automatize the selection process for a more relevant IDB. The proposed method of formulating queries for its further processing has an advantage in that it is done objectively and can be translated into an automatized software application without the direct intervention of a researcher or an expert.
The reported study was funded by RFBR and EISR according to the research project № 20-011-31460. 1